The number of impressions on the dashboard and via the outcome API are calculated in different ways.
The number of impressions on the participant/site dashboard (http://dashboard.trec-open-search.org/my/ssoar) shows the number of times a site has asked for a ranking (it counts the number of items in the feedback object returned by the db), which should correspond with the number of times it is shown to a user. Not all impressions contain actual feedback.
The "impressions" item in the object the outcome API returns filters on a lot more things, among other, whether actual feedback came in. (this is done by get_comparison()). Also, if the query type is "test", test query impressions/wins/losses are not yet shown during the test period.
Do we want to unify this and calculate API and dashboard impressions in the same way? or is this done deliberately